Skip to main content

Table 10 Meta-regression analyses for current smoking of any product (or cigarettes if any product not available) a

From: Systematic review with meta-analysis of the epidemiological evidence in the 1900s relating smoking to lung cancer

Characteristic Level All lung cancer Squamous Adeno
   Estimateb(SEc) pd Estimateb(SEc) pd Estimateb(SEc) pd
Constant   +1.566 (0.081)   +1.193 (0.239)   +0.816 (0.239)  
Sex Male Base <0.01 Base NS Base <0.001
  Female −0.204 (0.029)   −0.002 (0.088)   −0.440 (0.067)  
  Combined −0.043 (0.039)   +0.224 (0.459)   −0.612 (0.378)  
Location North America Base <0.001 Base <0.05 Base <0.001
  United Kingdom −0.345 (0.071)   No data   No data  
  Scandinavia −0.417 (0.066)   −0.177 (0.513)   −0.298 (0.320)  
  Other Europe −0.602 (0.057)   −0.440 (0.266)   −1.157 (0.178)  
  China −1.528 (0.097)   No data   No data  
  Japan −1.174 (0.056)   −1.174 (0.193)   −1.525 (0.136)  
  Other Asia −1.221 (0.139)   No data   No data  
  Other or multiregion −0.593 (0.088)   −1.221 (1.045)   −1.014 (0.754)  
Start year of study Before 1960 Base <0.001 Base <0.1 Base <0.05
  1960-69 +0.538 (0.063)   +1.374 (0.321)   +1.179 (0.231)  
  1970-79 +0.571 (0.058)   +0.896 (0.418)   +1.142 (0.258)  
  1980-89 +0.788 (0.051)   +1.276 (0.325)   +0.957 (0.224)  
  1990 or later +1.169 (0.089)   +2.817 (0.823)   +1.982 (0.575)  
Study type Case–control Base <0.1 Base NS Base NS
  Prospectivee +0.182 (0.047)   +0.404 (0.371)   +0.178 (0.224)  
Number of casesf 100-249 Base <0.1 Base NS Base NS
  250-499 +0.222 (0.071)   +0.420 (0.384)   −0.002 (0.195)  
  500-999 +0.266 (0.070)   +0.852 (0.400)   +0.646 (0.232)  
  1000+ +0.340 (0.062)   +0.803 (0.298)   +0.362 (0.149)  
Number of adjustment factors 0 Base NS Base <0.05 Base NS
  1 +0.002 (0.047)   +0.428 (0.277)   +0.195 (0.213)  
  2+ +0.069 (0.034)   −0.447 (0.175)   +0.283 (0.125)  
  1. a Based on the same data as for Table 8. See that table for further definition of RRs selected for analysis, and numbers of estimates of each characteristic level.
  2. b Estimates are of log RR. The predicted RR of a given estimate can be calculated by adding the constant to the values for the level of each characteristic applicable to the estimate (taking the value for a base level as zero) and taking the exponential of the result.
  3. c SE = standard error.
  4. d The p value is estimated from the drop in deviance from removing the characteristic from the fixed model using an F-test. It is expressed as p < 0.001, p < 0.01, p < 0.05, p < 0.1 or NS (p ≥ 0.1).
  5. e Or nested case–control or case-cohort in the case of 5 estimates for all lung cancer, 3 for squamous and 3 for adeno.
  6. f In the study as a whole.
  7. Note For the forward stepwise analyses, deviances, degrees of freedom and p values for inclusion in order of inclusion are as follows:
  8. All lung cancer: Constant 2669.216 on 194 d.f. (p < 0.001), location 1257.714 on 187 d.f. (p < 0.001), start year of study 918.536 on 183 d.f. (p < 0.001) and sex 869.479 on 181 d.f. (p < 0.01). Fixed model deviance 819.856 on 175 d.f.
  9. Squamous: Constant 288.688 on 40 d.f. (p < 0.001), location 194.753 on 36 d.f. (p < 0.01), start year of study 143.741 on 32 d.f. (p < 0.05) and number of adjustment variables 100.916 on 30 d.f. (p < 0.01). Fixed model deviance 86.049 on 24 d.f.
  10. Adeno: Constant 528.128 on 43 d.f. (p < 0.001), location 170.181 on 39 d.f. (p < 0.001), sex 128.397 on 37 d.f. (p < 0.01) and start year of study 89.243 on 33 d.f. (p < 0.05). Fixed model deviance 65.727 on 27 d.f.